Unfortunately not an option for us as for misc security reasons we can't permit our workstations to directly query the DNS server we currently forward to but thanks for the suggestion.
Regards
Allan
"Andrew Hodgson" <me3[ at ]privacy.net> wrote in message news:onfkf4lfh011abeakunu1n0f0is75726oo[ at ]news.giganews.com...
[Quoted Text] > On Tue, 14 Oct 2008 15:39:02 +0100, "Alan Sandal" > <alan.sandal[ at ]yahoo.co.uk> wrote: > >>Someone MUST have seen this before. >> >>I have several domains hosted on my Windows 2003 infrastructure and >>several >>domains listed for conditional forwarding. Name resolution for the >>authoritative domains and those specifically forwarded are working just >>fine. >> >>All Internet connectivity is via a proxy server in a DMZ and we have no >>requirement to allow workstations to resolve external DNS entries >>themselves - therefore there's no DNS connectivity from our internal DNS >>servers to the Internet. >> >> >>The problem comes when a machine queries for an address which isn't in >>either the hosted domains or in a domain for which a forwarder is >>specified >>e.g. www.microsoft.com. The internal servers can't resolve this, they have >>no forwarder specified to which they can forward it. >> >> >>The of course try to contact a root hints server. Now this is the >>problem - >>the timeout for this is several seconds and causes all machines making >>invalid queries to stop and wait for a timeout. In an ideal world (or at >>least an ideal network) there would be no incorrect queries but I've got >>lots of them and can't tackle the problem 'properly' by resolving the >>underlying problem. What I'd like to do is minimise the problem by sending >>a >>DNS failure msg immediately. I know: >> >>1. If I add a root domain to my servers I get an immediate DNS failure >>(good) but forwarding is disabled and I have several conditional >>forwarders. >> >>2. If I disable recursion for the server I get an immediate DNS failure >>(good) but forwarding is disabled and I have several conditional >>forwarders. > > This is what we did in this situation, but instead of using > conditional forwarders, we used stub zones which were AD integrated, > which created the necessary NS and A glue records for everything to > work fine. > > Andrew.
|